01:47:44 | Sammy | BZFlag: have you got my mail ? |
02:52:03 | Sammy | BZFlag: ? did U got mail from Sammy ? |
02:56:11 | BZFlag | I think so, yes. |
02:56:35 | Sammy | and " $ " ? |
02:56:50 | Sammy | you check ? |
10:23:27 | seletz | hi folks! |
10:23:54 | erikm | hello seletz |
10:23:59 | erikm | and morning, * |
10:27:22 | seletz | erikm: hi, and thanks for the driver diff. I looked at it, and found that i do quite the same on my board, except that i use russels last patches for pcmcia, so i more and more suspect a hw problem. sigh. |
10:27:36 | seletz | s/russel/russell/ |
10:28:46 | erikm | seletz: heh :) at times russell can be picky about his name :) |
10:29:24 | seletz | erikm: yes "russell, that's with two s and two l ..." |
10:29:47 | seletz | erikm: one never knows who's listening :)) |
10:32:59 | seletz | erikm: btw, did you read kernel traffic #150? The part 'bout a new scheduler that does 6M context switches per second? The numbers quite baffled me ... |
10:33:46 | seletz | detects heavy blob changes .... |
10:47:32 | seletz | debugging pcmcia |
10:49:04 | erikm | seletz: yeah, interesting thread. |
10:49:08 | erikm | SIGLUNCH |
10:49:14 | seletz | Maahlzeit! |
10:49:35 | seletz | :D |
11:40:27 | Sammy | goes home |
11:51:01 | erikm | re |
11:57:53 | seletz | erikm: on lart, have you been able to mount flash cards using ide-cs? |
12:00:15 | erikm | no, the PCMCIA doesn't work yet |
12:01:04 | seletz | erikm: actually what i want to know is: Does _someone_ actually use ide-cs to talk to flash cards on ARM? |
12:11:35 | erikm | seletz: yes, the ipaq people do |
12:22:04 | seletz | erikm: hmmm, i can dd from/to /dev/hda1, but mount hangs forever. |
12:22:16 | seletz | investigating |
14:54:11 | seletz | hmmm, russell's web site is down? |
14:56:06 | seletz | prpplague: hi! |
14:56:42 | prpplague | howdy seletz |
14:56:51 | prpplague | seletz: hows the biz going? |
14:57:17 | seletz | prpplague: fine, thx! Got _lots_ of work todo recentls ... |
14:57:28 | seletz | s/recents/recently/ |
14:58:17 | seletz | prpplague: atm i'm struggeling with ide flash cards and pcmcia on the PT Digital Board. |
14:58:30 | prpplague | seletz: ohh fun |
14:58:37 | seletz | prpplague: not really :( |
14:58:47 | prpplague | seletz: my point |
14:59:35 | seletz | prpplague: i can dd in/out from the flash through hdax but mount hangs the system. Get even an kernel oops in irq_check_lock(). sigh. |
15:00:04 | seletz | prpplague: not exactly reliable behavior, i'd say :/ |
15:00:37 | prpplague | seletz: definetly something causing a problem |
15:01:46 | seletz | prpplague: now i try to use ide.c as module and pass hda=slow hda=flash to it. Well, i'ts worth a try. |
15:02:23 | prpplague | seletz: have you compiled multi-mode ide access into the kernel |
15:02:24 | prpplague | ? |
15:02:52 | seletz | prpplague: not that i'm aware of. What's that? |
15:03:56 | prpplague | most flash type memory that emulates ide needs to have the multi-mode access configured under the ide functions |
15:04:07 | prpplague | standby let me get some more details for ya |
15:04:32 | seletz | prpplague: ok, thanks |
15:04:37 | seletz | waiting |
15:05:59 | prpplague | seletz: the config option is CONFIG_IDEDISK_MULTI_MODE , and its under the ATA/IDE options |
15:06:13 | seletz | one moment |
15:06:34 | prpplague | seletz: i ran into this problem using the Disk-On-Module units as well as CF cards |
15:06:43 | seletz | # CONFIG_IDEDISK_MULTI_MODE is not set |
15:06:46 | seletz | doh. |
15:07:06 | prpplague | seletz: i can't guarrante its all of your problem, but i'd bet its part of it |
15:07:21 | seletz | prpplague: i'll give it a try. Thanks a lot, anyway. |
15:07:37 | prpplague | no problem, i like to help when i can |
15:10:53 | seletz | configuring kernel |
15:11:47 | seletz | building new modules |
15:13:55 | seletz | flashing kernel |
15:22:53 | seletz | having insmod problems ... :( |
15:26:33 | prpplague | doh |
15:33:03 | seletz | prpplague: hmm, now ide-cs.o fails registering with the ide driver. Hmmm. |
15:33:11 | seletz | investigating |
15:34:21 | prpplague | seletz: thats odd |
15:35:07 | seletz | prpplague: i compiled everything as module: pcmcia subsystem, ide driver and ide card services module. |
15:35:37 | seletz | prpplague: before i had the ide driver 1) compiled into the kernel and 2) compiled w/o multi mode |
15:36:16 | seletz | prpplague: now i try to switch back to ide driver as module & w/o multi mode. Let's see. |
15:37:12 | prpplague | ok |
15:37:52 | seletz | prpplague: i want to be able to pass hda=flash to the ide driver ... |
15:38:03 | seletz | prpplague: w/o rebooting, that is |
15:39:03 | prpplague | seletz: this is odd, i've not had any problems with the multi-mode function, its solved all of my errors on the ide flash |
15:40:24 | seletz | prpplague: i'ts still possible that there's a odd hw bug. I'm able to dd on the raw dev, tough. That's really odd. |
15:59:02 | seletz | hmmm, still a no-go. I'm going to put the ide driver into the kernel. Hmm, something is wrong here. |
16:05:10 | seletz | reading Documentation/ide.txt :) |
16:12:47 | seletz | building kernel ... |
16:17:00 | prpplague | anyone seen these guys - http://www.ssv-embedded.de/ |
16:19:00 | seletz | btw, /,e checking out |
16:20:12 | prpplague | guess i should have really pointed to - http://www.linuxdevices.com/products/PD8550023611.html |
16:21:40 | seletz | cool |
16:23:24 | seletz | sadly enough the dont habe an sa1111 on it, and dont have some more infos on their homepage. |
16:24:11 | prpplague | seletz: hmm this looks cool - http://www.linuxdevices.com/products/PD5343937505.html |
16:26:38 | seletz | silly url names, dammit. |
16:29:47 | seletz | well, another arm based chip design. Cool is the on-chip frame buffer, It's a little tiny, tough. |
16:30:01 | erikm | looks |
16:30:25 | erikm | hmm, lineo support. /me guesses that BZFlag is involved :) |
16:30:31 | prpplague | lol |
16:30:44 | prpplague | hmm, i like the sound of this - http://www.linuxdevices.com/products/PD5343937505.html |
16:31:57 | prpplague | that might be nice for a newbie like me to get into the board level stuff |
16:33:56 | seletz | hmm, thats the same link, no? |
16:34:28 | seletz | oh, kernel is compiled :) |
16:35:40 | prpplague | seletz: doh |
16:37:31 | prpplague | seletz: too many people comming in and out of here today, didn't realize i'd already sent the link |
16:37:42 | prpplague | goes for more coffee |
16:38:51 | prpplague | erikm: what do think? think that sharp cpu would be easy to implement? |
16:39:13 | erikm | prpplague: after I'm done with the start.S cleanup it certainly will be |
16:40:13 | prpplague | erikm: sorry, i meant on the hardware level, i want to get to the point i can play around with my own design |
16:40:25 | erikm | prpplague: I'm currently splitting it up in an arch independent and arch dependent part |
16:40:49 | erikm | prpplague: oh, ok. well, the most difficuly part usually is the memory layout |
16:41:34 | prpplague | erikm: The LH7A400 provides glue-less interface to SRAM/SDRAM/Flash devices, MMC, PCMCIA/CF and Smart Cards. |
16:41:47 | seletz | erikm: oand to find someone to get a good pcb layout done, i guess... |
16:41:52 | prpplague | erikm: i wonder how true that is? |
16:42:37 | erikm | seletz: that's what I mean. the pcb layout for the memory interface is the most critical |
16:43:20 | seletz | erikm: i see. I guess that's where most of _my_ problems come from. |
16:44:50 | prpplague | erikm: btw, i had no doubt blob could boot on that cpu, after all, erikm is guru of bootloaders, lol |
16:45:19 | seletz | dupped |
16:45:21 | seletz | :) |
16:46:09 | prpplague | erikm: so what tends to be the issue on the memory layout? long traces creating timming issues? |
16:46:28 | erikm | prpplague: that, and traces of different length |
16:47:00 | erikm | prpplague: so D0 trace is longer than D15, and with a tight memory timing the D15 data comes too late at the CPU |
16:47:31 | prpplague | erikm: gotcha, so the latch for the data grabs bogus info |
16:47:43 | erikm | right |
16:48:27 | prpplague | man, i need to quite my job so i can get some work done! |
16:48:44 | erikm | the problem usually doesn't show up until you enable the cache and the CPU starts doing burst reads or writes |
16:49:10 | prpplague | erikm: sounds like you've had this problem before? on the lart? |
16:49:53 | erikm | prpplague: no, not on the LART. jdb hand routed the pcb for optimal memory speed. |
16:50:24 | seletz | _ouch_ |
16:50:43 | erikm | prpplague: but there have been problem reports about that on l-a-k |
16:50:54 | seletz | erikm: autorouter don't do good jobs, then? |
16:51:12 | erikm | seletz: no, autorouters are completely useless |
16:51:43 | erikm | seletz: one of our silly plans is to write an autorouter that actually works :) |
16:52:20 | seletz | erikm: :) |
16:52:47 | Russ | erikm: the new pcb lets you autoroute traces individualy or as groups |
16:52:50 | seletz | erikm: Tell me when youre done, so even ican create some pcb's :D |
16:52:53 | prpplague | erikm: one thing i've always wanted on a autorouter is a metric to weight the traces |
16:53:40 | prpplague | maybe there is one, i just haven't spent enough time playing around with them |
16:56:00 | seletz | suffering from another oops |
16:56:29 | erikm | o/~ do you suffer from long term memory loss? I don't remember... o/~ |
16:57:05 | prpplague | is suffering from an oops otherwise known as the boss |
16:57:24 | seletz | lol |
16:58:15 | seletz | nah, my pcmcia oopses the kernel. I now tend to call it a hw problem. |
17:10:13 | Russ | err... |
17:10:29 | erikm | hmm? |
17:11:36 | Russ | http://russ.dhs.org/files/tuxpatches/patch-2.4.13-ac8-rmk1-tux2.gz |
17:11:45 | Russ | grep this diff for rsrc_mgr.c |
17:12:38 | Russ | that hunk stops pcmcia from doing do_io_probe(), since the sa110 has no IO ports (but the kernel assumes it does since CONFIG_ISA is enabled) |
17:13:20 | erikm | shouldn't do_io_probe() probe in the PCMCIA IO space? |
17:13:38 | Russ | its always oopsed for me |
17:14:00 | Russ | changed it to #if defined(CONFIG_ISA) && !defined(CONFIG_ARCH_SA1100) instead of #ifdef CONFIG_ISA |
17:17:59 | seletz | erm, could someone explain that a little more, please? |
17:18:01 | Russ | at attempts to access x86 io space |
17:18:09 | Russ | and the addresses aren't mapped |
17:18:17 | Russ | maybe you could map them on your board and it would fix it |
17:18:33 | seletz | hmm, thus the "tries to free ressources not allocated" error |
17:19:15 | Russ | no, that isn't familiar |
17:19:22 | Russ | but backtrace the oops |
17:21:17 | seletz | hmm, it oopsed in check_irq_lock(), but it seems to completely random at times. the last one didn't have a vaild frame ptr, so no function adrs. |
17:21:23 | seletz | checking out diff |
17:27:18 | seletz | Russ: hmm, it's already fixed in 2.4.16-rmk2. There's a CONFIG_PCMCIA_PROBE and it's handled properly in Config.in |
17:27:45 | seletz | Russ: But thak you anyway! |
17:59:15 | seletz | Well, i'll call this a day. |
17:59:33 | seletz | bye, 'til tomorrow ... |